[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#1063140: mpg123: NMU diff for 64-bit time_t transition



Am Thu, 4 Apr 2024 09:36:37 +0200
schrieb Sebastian Ramacher <sramacher@debian.org>: 

> Now I get the following on arm{hf,el}:
> 
> --- debian/libmpg123-0.symbols (libmpg123-0_1.32.6~dev+20240403022201-1_armhf)
> +++ dpkg-gensymbolspYII3c	2024-04-03 09:52:12.863133592 +0000
> @@ -8,8 +8,8 @@
>   mpg123_current_decoder@Base 1.7.2
>   mpg123_decode@Base 1.6.2
>   mpg123_decode_frame64@Base 1.32.3
> - mpg123_decode_frame@Base 1.6.2
> - (arch-bits=32|arch=!x32)mpg123_decode_frame_32@Base 1.13.7
> +#MISSING: 1.32.6~dev+20240403022201-1# mpg123_decode_frame@Base 1.6.2
> +#MISSING: 1.32.6~dev+20240403022201-1# (arch-bits=32|arch=!x32)mpg123_decode_frame_32@Base 1.13.7
 
> From your explanation above, this is what I expected. The builds on 64
> bit architectures and i386 are unaffected.

Thanks for confirming.

> If you feel strongly about this, I think the best option would be to
> bump SONAME of the libraries upstream … 

I only feel that the subtle breakage with changed behaviour of the
still-existing symbol is to be avoided. My change for 1.32.6 does this.
The remaining symbols are still compatible with the pre-existing ABI
and upstream nothing changed, so an SONAME change from my side is not
really warranted.

This is a specific build configuration that Debian chooses. There are a
number of other options where you can disable library features and make
symbols disappear. So not that much changed from before, from the
upstream POV.

Unless you plan to add some SONAME change to all libraries affected by
the time_t/off_t change, I don't see why mpg123 libs should be singled
out. If this is just done by package name/version suffixes, then so be
it.

I'll release 1.32.6 now to enable a smoother transition.


Alrighty then,

Thomas


Reply to: